The regulation of CFTR by protein-protein interactions

The regulation of CFTR by protein-protein interactions

William R Thelin

A dissertation submitted to the faculty of the University of North Carolina at Chapel Hill in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the Department of Cell and Developmental Biology.

Chapel Hill 2006

Approved by:

Sharon L. Milgram, advisor Vytas A. Bankaitis, Ph.D. Michael D. Schaller, Ph.D.

iii ABSTRACT

William R. Thelin: The regulation of CFTR by protein-protein interactions

(Under the direction of Dr. Sharon L. Milgram)

Cystic fibrosis (CF) is an autosomal recessive disease resulting from the

mis-regulation of epithelial ion transport. CF is caused by mutations in the cystic fibrosis

transmembrane conductance regulator (CFTR), an apical membrane chloride channel

expressed in polarized epithelial cells. To identify factors that regulate CFTR activity, we

utilized biochemical and proteomics approaches to identify novel CFTR binding proteins.

We find that the C-terminus of CFTR directly interacts with the serine/threonine

phosphatase PP2A. PP2A is a heterotrimeric phosphatase composed of a catalytic subunit

and two divergent regulatory subunits (A and B), which mediate the cellular localization and

substrate specificity of the enzyme. By mass spectrometry, we identified the exact PP2A

regulatory subunits associated with CFTR as Aα and B’ε, and find that the B’ε subunit binds

CFTR directly. PP2A subunits localize to the apical surface of airway epithelia and PP2A

phosphatase activity co-purifies with CFTR in Calu-3 cells. In functional assays, PP2A

inhibition blocks the rundown of basal CFTR currents and increases channel activity in

excised patches of airway epithelia and in intact mouse jejunum. Moreover, PP2A inhibition

cells by a CFTR-dependent mechanism. Thus, PP2A is a relevant CFTR phosphatase in epithelial tissues and may be a clinically relevant drug target for CF.

Additionally, the N-terminus of CFTR directly interacts with two actin binding proteins, filamin A and filamin B. In polarized epithelial cells, filamins are highly localized to the sub-apical compartment where they likely interact with CFTR at or near the plasma membrane. We find that CFTR and filamins specifically interact by co-immunoprecipitation and that a disease-causing mutation in CFTR, serine 13 to phenylalanine (S13F), disrupts this interaction. Consistent with the loss of cytoskeletal anchorage, S13F CFTR displays decreased cell surface levels and less confinement at the plasma membrane relative to wild-type CFTR. Furthermore, S13F CFTR is more rapidly degraded compared to wild-wild-type CFTR which correlates with the accumulation of S13F CFTR in the lysosomes. Taken together, these data suggest the filamins regulate the cell surface stability and endocytic trafficking of CFTR.

v

ACKNOWLEDGEMENTS

I am thankful for all of the help and support that I have received during my graduate studies. First, I thank Sharon Milgram for providing me with a tremendous amount of scientific and professional guidance. I really appreciate how hard you have worked to help me achieve success inside and outside of the lab. I will miss the morning trips for coffee. I also thank Jack Stutts for always keeping his door open, many intellectual and experimental contributions, and helping me to navigate the CF center. I thank my committee members Vytas Bankaitis, Mike Schaller, Ken Hardin, and David Siderovski for challenging me to be a better scientist. I also thank Craig Dees for helping me to start on this path.

I thank the members of the Milgram lab for making our lab such a productive and fun place to work. I especially thank Caleb Hodson, Paul Barnes, and Mike Howell for good conversations. I also thank Gabby Haddock and Ashleigh Huggins who were a pleasure to work with.

I owe a great deal of thanks to my many collaborators who generously contributed their time, ideas and resources to these projects. At the UNC CF center, I thank Ric Boucher, Martina Gentzsch, Sylvia Kreda, Mehmet Kesimer, Rob Tarran, Mike Knowles, John Sheehan, Barbara Grubb, Ray Pickles, and Wanda O’Neil. I also thank Yun Chen, Jason Snyder, Mike Younger, and Jen Sallee for technical assistance, reagents, and good conversations.

I could not have made it this far without the encouragement of my family. Thank you for all of your support throughout this process.

TABLE OF CONTENTS

List of Tables ………..viii

List of Figures………...ix

List of Abbreviations ………...xi

Chapter 1: The physiology of CFTR………..…………1

Multiple protein interactions regulate the CFTR folding and stability……….….………..……..16

Interactions with coat proteins, adaptors, and vesicle fusion machinery regulate CFTR membrane trafficking…….….………..…….19

Cell surface CFTR is compartmentalized with the cAMP generating machinery and the cytoskeleton……….….………..….22

Chapter 2: A single step approach to analyze affinity purified protein complexes by mass spectrometry……….….………...26

Introduction……….…….………....26

Experimental Procedures……….…….………...28

Results……….…….………....30

Discussion……….……….………..…39

Chapter 3: CFTR is regulated by a direct interaction with the protein phosphatase PP2A ……….……….………..…42

Introduction……….………….………....42

Experimental Procedures ……….………….………...44

Results……….………….………....48

vii

Chapter 4: Filamins regulate the cell surface stability and endocytic

sorting of CFTR ………....76

Introduction………..76

Experimental Procedures ……….78

Results…..………86

Discussion ………..131

Chapter 5: Future directions ………...146

LIST OF TABLES

Table 1: A comparison of proteins identified by in-gel and in-solution approaches ………36

Table 2: Filamins regulates the biosynthesis, trafficking, and activity of numerous

transmembrane proteins ………..………..97

ix

LIST OF FIGURES

Figure 1.1: CFTR is an ABC transport protein ...……….…………..………..6

Figure 1.2: CFTR is activated downstream of adenosine receptor activation ….………...…10

Figure 1.3: CFTR channel gating reflects the balance of endogenous kinase and phosphatase activity ……….………...………13

Figure 2.1: Strategy to purify PDZ proteins associated with CFTR ...………….…..………32

Figure 2.2: The PDZ proteins identified by MS specifically associate with the CFTR C-terminus ……….………..………….…………38

Figure 3.1: Diverse subunits give rise to the specificity of PP2 ...51

Figure 3.2: MS analysis of PP2A subunits associated with the CFTR C-terminus ..………..53

Figure 3.3: Then PP2A heterotrimer co-purifies with CFTR ………..………...55

Figure 3.4: The B’ε subunit directly interacts with the CFTR C-terminus ……..…………..58

Figure 3.5: The PP2A B’ regulatory subunit localizes to the apical compartment of ciliated airway cells ....……….61

Figure 3.6: PP2A inhibitors functionally regulate CFTR ……..……….65

Figure 4.1: The N-terminus of CFTR is required for normal biogenesis and function ….….89

Figure 4.2: Filamins interact with the N-terminus of CFTR ……….….93

Figure 4.3: Filamins integrate cell signaling and mechanics ……….….95

Figure 4.4: The S13F mutation abolishes the interaction between full-length CFTR and filamin …….………....99

Figure 4.5: The N-terminus of CFTR associates with FLN repeats 1-4 ………...103

Figure 4.6: FLN-A co-precipitates with endogenous CFTR ………...…105

Figure 4.7: FLN-A localizes to the sub-apical compartment of polarized epithelia……...107

Figure 4.8: CFTR mutations affect filamin binding ………..………...112

Figure 4.10: S13F CFTR is expressed on the cell surface, but to a lesser extent

than wild-type ………...117

Figure 4.11: S13F reduces the plasma membrane confinement of CFTR .…………..…….120

Figure 4.12: S13F is prematurely trafficked to lysosomes where it is degraded ………..…125

Figure 4.13: CFTR surface expression is reduced in cells lacking FLN-A ………..…127

Figure 4.14: Competitive peptides decrease CFTR surface expression …...…..…………..129

Figure 4.15: The filamin binding motif of CFTR is similar to other filamin binding

proteins……….135

Figure 4.16: An N-terminal GFP fusion blocks filamin binding to CFTR …………..…….141

Figure 4.17: A model of CFTR polarized trafficking ………..….143

Figure 5.1: Calmodulin can also interact with the CFTR N-terminus …………..…………151

xi

LIST OF ABBREVIATIONS

A2BR Adenosine Receptor

ABP Actin binding protein

AMPK Adenosine monophosphate kinase

AP2 Adaptor protein 2

ASL Airway surface liquid

ATP Adenosine 5’-triphosphate

β2-AR Beta 2-aderenergic receptor

BHK Baby hamster kidney

CAL CFTR-associated ligand

cAMP 3'-5'-cyclic adenosine monophosphate

cGK-II cGMP dependent protein kinase II

CF Cystic fibrosis

CFTR Cystic fibrosis transmembrane conductance regulator

CHO Chinese hamster ovary

ER Endoplasmic reticulum

ERAD ER associated degradation

FLN Filamin

GCC Guanylyl cyclase C

GFP Green fluorescent protein

HSP Heat shock protein

HEK Human embryonic kidney

Isc Short circuit current

LC Liquid chromatography

MS Mass spectrometry

NBD Nucleotide binding domain

NHERF Sodium hydrogen exchange regulatory factor

R Domain Regulatory domain

PCL Pericilliary liquid

PDE Phosphodiesterase

PDZ Post-synaptic density 95, Discs large, ZO-1

PKA cAMP dependent protein kinase

PKC Protein kinase C

PP1 Protein phosphatase type 1

PP2 Protein phosphatase type 2

SPT Single particle tracking

SNARE Soluble N-ethylmaleimide-sensitive factor attachment protein receptors

SNX Sorting nexin

TCZ Transient confinement zone

Chapter 1: The physiology of CFTR

Epithelial cells provide a physical barrier which regulates the flow of substances

between an organism and its environment. Epithelia function to prevent bacteria and other

harmful substances from accessing the blood stream. Furthermore, epithelial cells must

balance their protective function with the selective uptake of nutrients and electrolytes. The

ability of epithelial cells to function as a barrier and transport tissue is derived from their

polarized nature. The apical (lumenal) plasma membrane of epithelial cells is physically

separated from the basolateral (serosal) membranes by tight junctions (Nelson et al., 1992;

Wollner and Nelson, 1992). Receptors, ion channels, and lipids are transported to selective

membrane domains, giving rise to the polarized distribution of signaling and transport

molecules. This segregation of cell surface proteins allows epithelia to achieve the vectorial

transport of water, ions, and other solutes.

The lungs are chronically exposed to pathogens and other noxious particles through

normal tidal breathing. Airway epithelial cells provide a first line of innate host defense by

regulating an elaborate clearance mechanism. In the lungs, the lumenal surface is covered by

a thin film known as the airway surface liquid (ASL) composed of a mix of gel and sol

(Widdicombe, 1989). The mucus gel is primarily composed of mucins, which are high

molecular weight glycoproteins secreted by goblet cells. The mucus layer traps bacteria and

inhaled particles preventing them from accessing the airway cells. The high carbohydrate

2

(Thornton et al., 1991). In addition to a purely passive role, antimicrobial peptides known as

defensins are secreted into the mucus where they are proposed to disrupt bacterial cell walls

by forming membrane pores (Chilvers and O'Callaghan, 2000). Mucus clearance is regulated

by ciliated epithelial cells which propel the mucus through the upper airways where it is

swallowed or expectorated. The normal function of airway cilia is dependent upon the

composition of periciliary fluid (PCL) underlying the mucus layer. The PCL provides a low

viscosity fluid which facilitates ciliary movement and is critical for mucociliary clearance

(Tarran et al., 2001).

Diseases which disrupt mucociliary clearance, such as cystic fibrosis (CF) or primary

ciliary dyskinesia, significantly increase susceptibility to life-threatening infections. CF is

caused by mutations in the gene encoding the cystic fibrosis transmembrane conducatance

regulator, an apical membrane chloride channel (Riordan et al., 1989). The loss of CFTR

function results in an imbalance of salt and water transport in airway epithelial cells. This

imbalance prevents the efficient clearance of mucus in the lungs ultimately leading to the

formation of thick mucus plugs in the lower airways (Quinton, 1999). Opportunistic

pathogens, such as Pseudomonas aeruginosa, colonize the mucus causing lung infections

(Kubesch et al., 1993). In CF patients, chronic lung infection is the primary cause of

morbidity.

Prior to the cloning of the gene encoding CFTR, many pioneering studies had

identified that CF was caused by the loss of a cAMP-dependent chloride conductance which

likewise altered the transport of other ions such as sodium and bicarbonate (Poulsen et al.,

1994; Smith et al., 1996; Stutts et al., 1997). In addition to a loss of apical chloride efflux,

prominent hypotheses which attempt to link the defect in ion transport to chronic lung

infections. The “compositional” hypothesis suggests that, in the absence of CFTR, epithelial

cell absorption of NaCl is decreased, thereby resulting in an increased NaCl concentration in

the ASL. At NaCl concentrations exceeding 100 mM, the function of the antimicrobial

defensins is impaired. Thus, the “compositional” hypothesis predicts that in the absence of

functional defensins, bacteria are able to colonize the mucus layer leading to infection.

Smith et al. found that bacterial did not efficiently grow when placed on the apical surface of

well-differentiated primary airway epithelial (WD-PAE) cells (Smith et al., 1996). However,

the mucus from CF cultures did not inhibit bacterial growth suggesting that the antimicrobial

properties were impaired. Consistent with the “compositional” hypothesis, adding water to

the apical surface of the CF cultures to decrease the salt concentration restored the ability of

the mucus to inhibit bacterial growth (Smith et al., 1996). A second hypothesis, known as

the “volume” hypothesis suggests that the ASL height, not ASL composition, is impaired in

CF airways. The “volume” hypothesis proposes that a dominant role of CFTR is to

negatively regulate the activity of the epithelial sodium channel (ENaC). Based on

observations that the activity of the epithelial sodium channel is significantly increased in CF

airways, this model predicts that isotonic salt transport results in a volume-depleted ALS

(Stutts et al., 1997). Furthermore, decreases in ASL volume affect mucociliary clearance

two-fold. First, the PCL height is reduced which impairs ciliary movements. Secondly, the

mucus layer becomes more viscous and is therefore more difficult to clear. Thus, mucus

accumulates in the lower airway which supports bacterial colonization. Recently, Mall et al.

provided strong support for the “volume” hypothesis by producing a CF-like phenotype in

4

measuring the salt concentrations of ASL derived from normal or CF airways would

distinguish between the “compositional” and “volume” models. However, the thin layer of

ASL has proven difficult to collect and analyze (Tarran et al., 2001). Alternatively, newer

approaches in which the ASL of WD-PAE cultures is labeled with fluorescent dextrans and

measured by confocal microscopy has provided a powerful means to study ASL volume.

Using this approach Tarran et al. found that the inhibition of CFTR results in a dramatic

decrease in ASL height (from 13 µM to 7 µM), while CFTR activation increases ASL height

(Tarran et al., 2005). These data provide strong evidence that CFTR activity is directly

coupled to the regulation of ASL volume.

In 1989, Riordan et al. reported the cloning of the gene encoding CFTR which led to

a rapid increase in our understanding of this disease (Riordan et al., 1989). Based on

sequence similarity, it was evident that CFTR is a member of the ATP-binding cassette

(ABC) transporter superfamily. ABC transporters include a large number of proteins (48

ABC transporter genes in humans alone) found in both prokaryotic and eukaryotic organisms

which transport a diversity of substrates including ions, peptides, lipids, and other solutes.

CFTR, like other ABC transporters, is topologically composed of two membrane-spanning

domains (MSD), which form the membrane pore, and two nucleotide binding domains

(NBDs) (Riordan et al., 1989). The NBDs contain sequences predicted to bind and

hydrolyze MgATP, such as the Walker A, Walker B, and LSGGQ motifs. In addition, CFTR

is dually glycosylated on asparagine residues 894 and 900 in the 4th extracellular loop, which

may have a role in CFTR folding and degradation (Cheng et al., 1990) (Figure 1.1).

Since the identification of the gene encoding CFTR, more than 1,000 different

Figure 1.1. CFTR is an ABC transport protein

functionally categorized by defects in early biosynthesis and folding (Class I, II, and V) or

chloride channel activity (Class III and IV) (Welsh and Smith, 1993; Zielenski and Tsui,

1995). Class I mutations are caused by nonsense mutations which prematurely truncate the

protein. Class II mutations interfere with the folding of CFTR in the endoplasmic reticulum

(ER). As a result, CFTR is retained in the ER where it is eventually targeted for degradation

by the ER-associated degradation pathway (ERAD). Class II mutations are the most

common and include the ∆F508 (deletion of phenylalanine 508) mutation found in ~70% of

all CF patients. Class III and IV mutations affect channel regulation at the cell surface. For

example, the G551D, a class III mutation, occurs in the LSGGQ motif in NBD1 and severely

impairs channel gating (Delaney et al., 1996; Oceandy et al., 2003; Oceandy et al., 2002).

Class IV mutations, such as R117H, result in altered channel conductance. Finally, class V

mutations result in reduced CFTR mRNA and, therefore, reduced protein (Zielenski and

Tsui, 1995). More recently, Haardt et al. proposed a new class of CFTR mutations (Class

VI) which cause the premature degradation of CFTR in a post-ER compartment (Haardt et

al., 1999). The class VI mutations are based on the observation that CFTR proteins lacking

at least the last 80 C-terminal amino acids are rapidly retrieved from the cell surface and

degraded through a ubiquitin-dependent pathway.

Initial studies in cells which lacked a CFTR-like activity including Chinese Hamster

ovary (CHO) (Anderson et al., 1991b), HeLa (Gregory et al., 1990), NIH-3T3 (Anderson et

al., 1991b), and Xenopus oocytes (Bear et al., 1991) demonstrated that the expression of

CFTR produced a unique cAMP-dependent chloride conductance. Furthermore, in excised

membrane patches, the chloride channel activity of CFTR was shown to require the activity

8

other ABC transporters, ATP hydrolysis at the NBDs provides the energy to actively

transport substances across the plasma membrane. Initially, it was not clear why CFTR

required the NBDs as chloride is passively transported through CFTR down a favorable

concentration gradient. However, it is now clear that the hydrolysis of ATP by the NBDs

provide the driving force for channel opening (Baukrowitz et al., 1994). The requirement for

PKA activity comes from the unique CFTR regulatory domain (R domain). The CFTR R

domain is a 272 amino acid stretch which separates NBD1 from MSD2. The R domain

contains multiple di-basic PKA consensus phosphorylation motifs. As such, PKA can

phosphorylate the CFTR R domain at least 10 times (Seibert et al., 1995). The

phosphorylation of the R domain increases the rate of ATP hydrolysis by the NBDs and

thereby increases channel opening (Howell et al., 2004) (Figure 1.2). The phosphorylation of

the R domain is thought to induce a conformational change in CFTR which may promote

interactions between NBD1 and NBD2 (Howell et al., 2004). In studies designed to examine

the phosphorylation-dependent structural changes in R domain structure, Ostedgaard et al.

found that the R domain is relatively unstructured in solution (Ostedgaard et al., 2000).

Presently, it is not clear whether the R domain functions as an activation domain or inhibitory

domain. The addition of exogenous R domain proteins blocked CFTR channel activity in

patches, suggesting an inhibitory function (Ma et al., 1996). However, the R domain likely

functions to stimulate channel activity as its phosphorylation potentiates CFTR activity to

higher levels than CFTR mutants which do not require phosphorylation for gating

(Baldursson et al., 2001). While the importance of the R domain with respect to CFTR

channel activity is well appreciated, the mechanism by which the R domain functions is not

Figure 1.2 CFTR is activated downstream of adenosine receptor activation

approaches are addressing how the R domain physically associates with the NBDs and how

R domain phosphorylation influences these interactions.

The channel activity of CFTR is tightly controlled by a balance of kinase and

phosphatase activity. CFTR is predominantly activated by PKA phosphorylation. However,

other kinases have also been shown to regulate CFTR including Ca2+-dependent and

independent PKC isoforms (Chappe et al., 2003; Dahan et al., 2001; Liedtke et al., 2002), the

cGMP dependent protein kinase II (cGKII) (Dahan et al., 2001), and the tyrosine kinase

c-Src (Fischer and Machen, 1996). When added exogenously to patch preparations, PKC

stimulates CFTR activity. However, the magnitude of CFTR activation by PKC alone is only

~15% of that by PKA (Chang et al., 1993). More importantly, Jia et al. showed that PKC

phosphorylation was critical for the potentiation of CFTR channel activity by PKA (Jia et al.,

1997). Based on these studies, they proposed a model whereby the constitutive

phosphorylation of CFTR by PKC is required for PKA dependent activation of the channel.

In addition, cGKII phosphorylation of CFTR activates the channel comparably to PKA

activation (Dahan et al., 2001). However, the kinetics of cGKII phosphorylation is slower

than PKA which has been predicted to reflect the initial requirement of membrane

association for activated cGKII. The addition of the tyrosine kinase c-Src in excised patches

increases CFTR gating; however, the physiological significance of Src-mediated CFTR

regulation is not clear (Fischer and Machen, 1996).

The signaling pathways which activate CFTR in vivo are tissue dependent. In the

airways, CFTR is predominantly activated by adenosine receptor (A2B) signaling, a Gs

coupled G-protein coupled receptor (Eidelman et al., 1992). Activation of the A2B receptor

12

Figure 1.3 CFTR channel gating reflects the balance of endogenous kinase and phosphatase activity

14

activated in response to the elevated cAMP levels, leading to CFTR phosphorylation and

increased channel gating (Figure 1.3). CFTR can also be potently activated by the B2

-adrenergic receptor (β2AR) similarly to the A2B receptor. The β2AR can activate CFTR in

tissues such as sweat ducts. However, it is not clear whether the β2AR is a physiologically

relevant activator of CFTR in the airways. In the intestines, CFTR is alternatively activated

through cyclic GMP signaling pathways. Guanylate cyclase C (GCC) is an integral

membrane guanylate cyclase expressed in the gut. GCC converts GTP into cGMP, in

response to ligands such as guanylin or uroguanylin (Li et al., 1995b). CFTR is activated

downstream of GCC activation by the cooperative actions of cGKII and PKA (Tien et al.,

1994). In addition to its endogenous ligands, GCC is activated by heat stable enterotoxins

(STa) elaborated by strains of pathogenic E. coli (Schultz et al., 1990). STa activation of

GCC results in apical chloride efflux through CFTR, as well as, the inhibition of sodium

transport through the sodium hydrogen exchange regulatory factor 3 (NHE-3). The net result

of this signaling pathway is increased salt in the intestinal lumen. This likewise provides a

driving force for water movement into the lumen resulting in secrectory diarrhea (Zhang et

al., 1999). In less developed countries, secretory diarrhea is a life-threatening health concern.

While the activation of CFTR has been extensively studied, much less is known about

channel deactivation. Once it was clear that CFTR activity required kinase activity,

phosphatases were obvious candidates for negative channel regulators. Work from many

labs suggests that multiple phosphatases including PP2A, PP2B, PP2C, and alkaline

phosphatase are involved in the deactivation of CFTR (Luo et al., 1998; Reddy and Quinton,

1996). Based on studies using exogenous phosphatases, PP2A and PP2C are the most

to patch preparations, purified PP2A and PP2C both inactivate maximally phosphorylated

CFTR (Luo et al., 1998). However, PP2C altered channel openings, while PP2A shortened

CFTR burst duration. These data suggest that PP2A and PP2C distinctly regulate CFTR

channel activity. Cell permeable phosphatase inhibitors have been used to study CFTR

deactivation in intact cells. In human sweat ducts, cardiac myocytes, 3T3 fibroblasts, and

Hi-5 insect cells, inhibitors of PP2A, such as okadaic acid and calyculin A, increase CFTR

channel activity (Becq et al., 1994; Luo et al., 1998; Yang et al., 1997). In contrast, PP2A

inhibitors were observed to have little effect on the rapid deactivation of maximally

phosphorylated CFTR in T84 intestinal epithelial cells and human airway epithelial cells

(Travis et al., 1997; Zhu et al., 1999). However, the role of PP2A in the regulation of CFTR

in polarized epithelia has not been well characterized. In baby hamster kidney cells (BHK),

PP2C co-immunoprecipitates with exogenous CFTR (Zhu et al., 1999). Additionally, PP2C

overexpression in Fischer rat thyroid cells (FRT) decreases CFTR chloride conductance

(Travis et al., 1997). However, the ability of endogenous PP2C to regulate CFTR in native

epithelial tissues is unclear as no PP2C inhibitors have been identified. Furthermore, studies

in NIH-3T3 fibroblasts suggest that the inhibition of PP2B also stimulates CFTR channel

activity (Fischer et al., 1998). However, it is not clear if PP2B is a relevant CFTR

phosphatase in vivo.

To date, many signaling molecules able to regulate CFTR have been identified.

However, the ability of many of these protein kinases and phosphatases to modulate CFTR

activity has been studied in heterologous expression systems, which may not accurately

recapitulate the epithelial-specific regulation of CFTR. In polarized epithelia, CFTR is

16

into protein complexes with regulatory factors. Furthermore, it is now appreciated that all

aspects of CFTR biology are regulated by direct and indirect protein interactions. The study

of CFTR protein interactions has provided a great deal of insight into the processes which

regulate CFTR biogenesis, trafficking, and channel gating. The remainder of this chapter

recounts much of what is known regarding the proteins that directly interact with CFTR.

Multiple protein interactions regulate the CFTR folding and stability

Genetic and biochemical screens for CFTR-interacting proteins have provided

significant insights into the chaperone proteins which associate with CFTR including heat

shock protein 40 (Hsp40), Hsp70, Hsp90, cysteine string protein (CSP), and calnexin

(Farinha et al., 2002; Loo et al., 1998b; Meacham et al., 1999; Pind et al., 1994; Yang et al.,

1993; Zhang et al., 2002). The majority of CFTR mutations, including the ∆F508 mutation,

result in channel misfolding, highlighting the importance of chaperones in CFTR biogenesis.

Chaperones dually function to promote the folding of nascent protein chains and, conversely,

the degradation of mis-folded proteins. During protein folding, chaperones such as Hsp70

interact with hydrophobic regions which are generally buried within the native protein. By

masking exposed hydrophobic amino acid stretches, chaperone proteins prevent protein

aggregation and coordinately facilitate proper folding. In Chinese hamster ovary (CHO) and

baby hamster kidney (BHK) cells, the overexpression of the human Hsp40 protein, Hdj-1,

and Hsp70 stabilizes the ER-form of wild-type CFTR, but not ∆F508 CFTR (Farinha et al.,

2002; Meacham et al., 1999; Yang et al., 1993). Loo et al. demonstrated that, in CHO and

BHK cells stably expressing wild-type CFTR, treatment with geldanamycin or herbimycin A

the loss of Hsp90 binding accelerated the degradation of the ER-form of CFTR via the

proteasome. Alternatively, overexpression of CSP with wild-type CFTR in HEK-293 cells

stabilized and retained CFTR in the ER (Zhang et al., 2002). Thus, the modulation of

chaperone levels impacts the folding and ER-exit of CFTR.

CFTR mutants that fail to achieve a stable conformation are retained in the ER and

ultimately degraded via the ER-associated degradation pathway (ERAD). CFTR degradation

by ERAD depends upon the retention of non-native CFTR in the ER followed by

proteosomal targeting. Several mechanisms have been proposed to regulate the ER retention

of CFTR which include sorting motifs and the glycosylation state of CFTR. CFTR contains

four arginine-based RXR motifs which have been shown to function as ER retention signals

(Chang et al., 1999). The introduction of arginine to lysine mutations in all four motifs

(RXR to RXK) in the context of ∆F508 CFTR allowed ~90% of the mutant protein to escape

the ER. However, the importance of these motifs for normal CFTR biogenesis is not

understood. In addition, the ER resident chaperone calnexin is implicated in the

ER-retention of CFTR (Pind et al., 1994). Calnexins interact with the N-linked oligosaccharide

chains of CFTR in a calcium-dependent manner. The processing of the glycosylation sites of

native CFTR by glucosidases results in the disassociation of CFTR and calnexin. However,

for non-native CFTR, the glycosylation sites are not fully processed, leading to a prolonged

interaction with calnexin and ER-retention. Once retained in the ER, chaperones such as

Hsp70 can recruit ubiquitination machinery including the E2 ubiquitin ligase UbcH5a

(Younger et al., 2004) and the E3 ubiquin ligase CHIP (Meacham et al., 2001). The addition

18

One caveat to the majority of CFTR-chaperone studies is that these interactions have

predominantly been studied using heterologous expression systems where CFTR matures

inefficiently (Ward and Kopito, 1994). In most heterologous systems, only ~30% of

wild-type CFTR exits the ER. However, in epithelial cell lines where CFTR is endogenously

expressed, CFTR is efficiently processed with nearly 100% of newly synthesized protein

reaching a post-ER compartment (Varga et al., 2004). The efficient processing of CFTR in

epithelia, but not other cell types, may reflect lower expression levels in epithelia and/or the

appropriate complement of chaperones and other regulatory proteins necessary for efficient

biogenesis. It will be important in future studies to characterize the interactions between

CFTR and chaperones in relevant epithelial cells.

The observation that most disease-causing mutations in CFTR, including the ∆F508

mutation found in ~70% of CF patients, are retained and degraded in the ER has significantly

increased the interest in understanding the processes which regulate CFTR biosynthesis.

Maneuvers such as using chemical chaperones (such as glycerol) or reduced temperatures

(<30°C) can facilitate the ER-exit of non-native CFTR where to allow its function at the

plasma membrane, albeit at a lower capacity (Brown et al., 1996; Denning et al., 1992).

Thus, rescuing non-native CFTR is currently being investigated as an avenue for CF

therapeutics. Recently, inhibitors of the calcium-adenosine triphosphate (SERCA) pump,

such as curcumin and thapsigargin, have received a great deal of attention as they have been

proposed to increase the export of ∆F508 CFTR from the ER (Egan et al., 2002; Egan et al.,

2004). Prolonged treatment with SERCA pump inhibitors depletes luminal calcium from the

ER and presumably causes the disassociation of calnexin from non-native CFTR. While

∆F508 mouse model; other groups have been unable to validate these findings (Grubb et al.,

2005; Song et al., 2004).

Despite the controversies over SERCA pump inhibitors, the rescue of mutant CFTR

proteins such as ∆F508 is still under investigation as a potential method to treat CF.

However, recent studies have suggested that this approach is complicated by the fact that

temperature-rescued ∆F508 CFTR exhibits altered channel gating and decreased stability at

the cell surface (Clarke et al., 2004; Egan et al., 2002; Egan et al., 2004; Gentzsch et al.,

2004; Swiatecka-Urban et al., 2005a). Thus, the identification of factors which regulate the

membrane trafficking and channel activity of CFTR will not only further our understanding

of normal CFTR physiology, but may provide insights into therapeutic approaches which

could be useful for augmenting the activity of mutant channels.

Interactions with coat proteins, adaptors, and vesicle fusion machinery regulate CFTR

membrane trafficking

The mechanisms which regulate the trafficking of CFTR through the early secretory

pathway are not well understood. CFTR, like other ER cargo destined for the Golgi, is

concentrated at ER exit sites where it is packed into COP-II coated vesicles. Inhibition of

COP-II coat assembly by the overexpression of dominant negative Sar1 GTPases causes

CFTR to accumulate in the ER (Bannykh et al., 2000). Wang et al. suggested that the

packaging of CFTR into COPII vesicles in mediated by a direct interaction between a

di-acidic (D-A-D) motif found in CFTR NBD1 with two components of the COP-II coat

(Sec23/Sec24) (Wang et al., 2004). The mutation of the second aspartic acid in the di-acidic

20

maturation. However, these studies do not address whether the defects associated with the

D567A mutation are specific to the COPII interaction or simply cause a global folding defect

in CFTR.

Little is known about the trafficking of CFTR to the Golgi complex following ER

export. Studies which follow CFTR maturation often rely on the observation that the

core-glycosylated, ER form of CFTR (band B) migrates with an apparent molecular weight of 160

kD. As CFTR is trafficked through the cis and medial Golgi, the N-glycan is processed to a

complex oligosaccharide, decreasing the electrophoretic mobility to 180 kD (band C) (Cheng

et al., 1990). The maturation of wild-type CFTR is not affected by the expression of

dominant negative Arf1, Rab1a, or Rab2 GTPases, which block conventional ER-to-Golgi

trafficking (Yoo et al., 2002). However, the overexpression of syntaxin 13, a target SNARE

expressed on late endosomes, blocked CFTR maturation. These data suggest that the

exocytic trafficking of CFTR follows a non-conventional pathway which involves the sorting

of CFTR from the ER to the late Golgi or endosomal compartments.

In the Golgi, CFTR interact with the PDZ domain-containing protein the

“CFTR-associated ligand” (CAL) (Cheng et al., 2002). CAL is composed of two N-terminal

coiled-coil repeats followed by a single PDZ domain (these domains are discussed in detail below).

CAL is broadly expressed across tissues and predominantly localizes to the trans-Golgi

network (TGN). Overexpression of CAL reduces the cell surface pool of CFTR by

decreasing both the half-life of CFTR and the insertion of CFTR into the plasma membrane

(Cheng et al., 2002). Subsequent studies have revealed that CAL overexpression targets

mature CFTR for lysosomal degradation (Cheng et al., 2005). Furthermore, the expression

surface expression (Cheng et al., 2002). These data have led to a model whereby PDZ

proteins “hand-off” CFTR in the Golgi or endosomal compartment to regulate the surface

expression of CFTR. However, the importance of CAL in the regulation of CFTR has not

yet been studied in relevant epithelial cells.

The surface expression of CFTR is also regulated by multiple soluble

N-ethylmaleimide-sensitive factor attachment protein receptors (SNAREs) which are

components of the vesicular fusion machinery. The V-SNARE syntaxin 1A directly binds

the N-terminus of CFTR to regulate CFTR trafficking in Xenopus oocytes (Peters et al.,

1999). Furthermore, this interaction also down regulates CFTR channel activity by reducing

the open probability of CFTR at the apical cell surface (Naren et al., 2000; Naren et al., 1997;

Naren et al., 1998). Syntaxin 1A may also cooperatively regulate CFTR with the t-SNARE

SNAP-23 (synaptosome-associated protein of 23 kDa) (Cormet-Boyaka et al., 2002). Like

syntaxin 1A, SNAP-23 negatively regulates channel gating. In addition, syntaxin 1a

potentiates the SNAP-23-CFTR interaction, suggesting that these cognate t-SNARES may

coordinately regulate CFTR as a heterodimeric complex. Moreover, the binding of Munc-18

to syntaxin 1A reverses the inhibitory effect on CFTR open probability, adding another layer

of complexity to this regulatory interaction (Naren et al., 1997). Syntaxin 8 may also

regulate CFTR function by inhibiting channel activity and by decreasing the amount of

protein on the plasma membrane, but it is unclear whether the protein directly binds CFTR or

functions in this capacity in polarized epithelial cells (Bilan et al., 2004).

Once CFTR reaches the cell surface it is rapidly internalized by clathrin-coated

vesicles. The clathrin adaptor AP2 directly binds GST-CFTR C-terminal fusion proteins in

tyrosine-22

based internalization motifs (Hu et al., 2001; Weixel and Bradbury, 2000). Once it is cleared

from the cell surface, CFTR can be rapidly recycled back to the plasma membrane (Gentzsch

et al., 2004; Picciano et al., 2003). Although the molecular mechanisms are poorly

understood, especially in polarized cells, a number of proteins including RME1 (Picciano et

al., 2003), Rab proteins (Gentzsch et al., 2004), myosin VI (Swiatecka-Urban et al., 2004),

and PDZ proteins (Swiatecka-Urban et al., 2002) have been implicated at various stages of

the retrieval and recycling process. As mentioned previously, there is a growing interest in

the development of therapeutic agents that facilitate the ER-exit of ∆F508 CFTR. Numerous

studies have demonstrated that these rescued ∆F508 proteins are trafficked to the cell surface

and have chloride channel activity (Denning et al., 1992). However, more recently, several

studies have shown that resuced ∆F508 is rapidly internalized from the cell surface and

inefficiently recycled (Gentzsch et al., 2004; Swiatecka-Urban et al., 2005a). Thus, impaired

trafficking may complicate ∆F508 rescue as a therapeutic approach. These observations

highlight the importance of elucidating the mechanisms which regulate the cell surface

stability and recycling of CFTR.

Cell surface CFTR is compartmentalized with the cAMP generating machinery and the

cytoskeleton

At the apical cell surface, CFTR is compartmentalized with regulatory factors which

modulate CFTR channel activity. Our previous work and the results of others are consistent

with the membrane localization of the cAMP signaling machinery that regulates CFTR. In

excised membrane patches, endogenous, membrane-associated PKA activity activates CFTR

is activated by receptors (Huang et al., 2001; Huang et al., 2000). Our lab previously used

small peptides to specifically disrupt protein interactions in Calu-3 cells and further showed

that PKAII is targeted to the plasma membrane by association with A-kinase anchoring

proteins (AKAPs) (Huang et al., 2000), one of which may be the cytoskeletal-associated

protein ezrin (Sun et al., 2000a). Our lab examined the activation of CFTR by adenosine

using excised and cell-attached membrane patches from Calu-3 cells. The addition of

adenosine to excised membrane patches potently stimulates CFTR channel activity (Huang et

al., 2001). These data demonstrate that the adenosine receptor, G-proteins, adenylate

cyclase, and PKA are all compartmentalized with CFTR within the patch preparations.

Furthermore, in cell attached patches of Calu-3 cells, the addition of adenosine stimulated

CFTR activity when it was added to the pipette solution (Huang et al., 2001). The local

activation of CFTR by adenosine receptor signaling is mediated by phosphodiesterases

(PDE), likely a member of the PDE4D family, which form a cAMP diffusion barrier around

the channel (Barnes et al., 2005). In a yeast two-hybrid assay, Hallows et al. identified the

AMP regulated kinase as a direct binding partner of the CFTR C-terminus (Hallows et al.,

2000). AMP kinases phosphorylate CFTR in vitro and inhibit CFTR-mediated chloride

conductance in Xenopus oocytes and Calu-3 cells.

At the apical cell surface, CFTR interacts with the highly related multi-PDZ proteins,

NHERF-1 (EBP50), NHERF-2 (E3KARP), and NHERF-3 (PDZK1 or CAP-70) (Short et al.,

1998; Sun et al., 2000b; Wang et al., 2000). Both NHERF-1 and -2 are composed of two

tandem PDZ domains and a COOH-terminal Ezrin/Radixin/Moesin (ERM) binding domain

(Reczek et al., 1997). CFTR binds to the first PDZ domain of NHERFs-1 and -2 which are

24

proposed to stabilize CFTR at the apical cell surface. As such, GFP-CFTR lacking the PDZ

binding motif exhibits a 60% increase in diffusional mobility compared to wild-type

GFP-CFTR (Haggie et al., 2002; Haggie et al., 2004; Moyer et al., 1999; Moyer et al., 2000).

Additionally, ezrin can interact with the regulatory subunit of PKA, scaffolding this kinase in

close proximity to the channel (Sun et al., 2000a). Naren et al. demonstrated that NHERF-1

can exist in a macromolecular complex simultaneously with CFTR and the β2 adrenergic

receptor, an upstream activator of CFTR (Naren et al., 2003). Furthermore, Wang et al.

demonstrated that PDZK1, facilitates CFTR dimerization by binding to two channels

simultaneously, thereby potentiating CFTR activity (Wang et al., 2000).

More controversial is the role of PDZ proteins in the establishment of the apical

polarity of CFTR. Using GFP-tagged CFTR constructs lacking the PDZ binding motif

(CFTR-∆TRL), Moyer et al. found that CFTR mis-localized to the lateral membranes (Moyer

et al., 1999). More recent studies from this group suggest that CFTR is trafficked to the cell

surface in an unpolarized fashion (Moyer et al., 2000). However, CFTR is concentrated at

the apical cell surface via an efficient recycling mechanism which is dependent on binding to

PDZ proteins. In contrast, three other groups have found that the apical trafficking and

retention of CFTR was not dependent upon the PDZ binding motif (Benharouga et al., 2003;

Milewski et al., 2001; Ostedgaard et al., 2003). These studies utilized either untagged or

HA-tagged CFTR-∆TRL expressed in epithelial cells. One interpretation of these data is that

the presence of the large N-terminal GFP-tag, in combination with the deletion of the PDZ

binding motif, contributes to the loss of apical polarity. However, these studies illustrate that

Numerous regulatory factors have been identified for CFTR; however, it is not clear

which ones are relevant in epithelial tissues or how they are organized. Using biochemical

and proteomics techniques described in chapter 2, we identified two novel CFTR binding

proteins, the B’ε subunit of the protein phosphatase PP2A and the cytoskeletal protein

filamin. We find that the B’ε subunit interacts with the highly conserved C-terminus of

CFTR, and this interaction targets a unique PP2A enzyme to CFTR. Furthermore, our data

demonstrate that PP2A functionally regulates CFTR in airway epithelia and may be a

relevant drug target for CF. In a similar screen, we find that filamins associate with the

N-terminus of CFTR. The interaction with filamins regulates the cell surface stability and

endocytic trafficking of CFTR. Taken together, our research highlights the power of using

proteomics and mass spectrometry to identify functionally relevant protein-protein

26

Chapter 2: A single step approach to analyze affinity purified protein complexes by

mass spectrometry

Introduction

The adaptation of mass spectrometry to the study of proteins has provided a powerful

tool to analyze protein-protein interactions. In combination with classic biochemical

approaches such as affinity purification or immunoprecipitation, MS provides a highly

sensitive and unbiased method to identify the constituents of protein complexes. This basic

approach has been extensively applied to studies ranging from the identification of individual

protein interactions to the characterization of entire interactomes. However, relevant

biological samples are often not compatible with MS analysis due to the presence of

contaminating substances. These contaminants include salts and detergents, which are

necessary for biochemical assays, but interfere with protein digestion and the quality of MS

data (Parker et al., 2005). In addition, highly abundant proteins can mask and suppress the

signals of less abundant proteins (Annesley, 2003). Depending on the method used,

contaminants include the fusion proteins (such as GST) or antibodies used to purify protein

complexes in pulldown or immunoprecipitation experiments. Furthermore, proteins which

interact with fusion proteins or antibodies non-specifically increase the sample complexity,

making it more difficult to identify the relevant proteins by MS.

To minimize sample contaminants, a purification step is often utilized prior to MS

analysis. Most commonly, proteins are separated by SDS-PAGE, visualized by coomassie

in-gel digestion effectively eliminates many contaminants such as salts and detergents, while

protein contaminants are resolved into discrete bands away from the proteins of interest. One

drawback to SDS-PAGE separation is that proteins must be detected by coomassie blue or

other detection methods, which may not be sensitive enough to detect less abundant proteins

(Loiselle et al., 2005). Furthermore, peptide fragments are not fully recovered from the gel,

often times less than 50%, resulting in lower sequence coverages (Alaiya et al., 2001;

Oppermann et al., 2000). Alternatively, proteins or peptides can be purified in-solution using

additional chromatography steps or dialysis to minimize high concentrations of salts and

detergents, but not protein contaminants. Despite their effectiveness to “clean-up” sample

contaminants, each of these methods results in significant sample loss, which is particularly

problematic for studying low abundance proteins in the sample.

In this chapter, we detail the development of a new method to prepare and analyze

protein complexes by mass spectrometry. Our goal was to develop a biochemical approach

to study protein interactions which is MS compatible without requiring additional

purification steps. We reasoned that this would allow us to achieve higher quality MS data

and identify more proteins in our samples. To accomplish this we designed a procedure

which allowed us to purify protein complexes while minimizing sample contamination. The

combination of our approach with MS analysis provides a powerful means to analyze

protein-protein interactions. In the subsequent chapters, we describe the regulation of CFTR

28 Experimental Procedures

Lysate preparation and protein purification

Mouse kidneys were homogenized in 10 volumes of binding buffer 150 (BB150: 50

mM Tris pH 7.6, 150mM NaCl, 0.2% CHAPS, 10mM EDTA, Roche complete protease

inhibitor cocktail, and 1 mM PMSF) by physical disruption using a tissue homogenizer on

ice. Lysates were tumbled at 4°C for 1 hour. Insoluble material was removed by

ultracentrifugation at 100,000xg for 1 hour. Protein concentrations of cell lysates were

determined by BCA assay (Pierce) and the final protein concentration was adjusted to 1

mg/ml. Lysate were pre-cleared of proteins endogenously coupled to biotin by tumbling with

streptadvidin agarose beads for 1 hour at 4°C at a ratio of 20 µl of settled beads per 1 ml

lysate. The beads were removed by centrifugation at 3,000xg for 10min. Pre-cleared lysates

were tumbled with the CFTR affinity matrices (20 nmoles of CFTR peptides immobilized to

100 µl of streptavidin agarose beads) for 4 hours at 4°C. Lysates were washed 5 times in 10

mls of BB150 without protease inhibitors for 5 minutes per wash. As a final wash step to

remove excess salt and detergent, samples were washed briefly in a low salt buffer (10 mM

sodium phosphate, pH 7.5 with 75 mM NaCl). Excess buffer was aspirated using a gel

loading pipette tip.

Sample elution

To elute proteins associated with the CFTR affinity matrix, the beads were incubated

in 1 volume (100µl) of 10% formic acid on ice for 5 minutes. ddH20 was then added to a

final volume of 1 ml per sample (adjusting the final concentration of formic acid to 1%). The

decreasing the formic acid concentration so as not to introduce covalent modifications or

hydrolyze the proteins. Finally, the beads were pelleted and the supernatant was collected,

frozen, and lyophilized.

MS Analysis

The lyophilized samples were processed and analyzed by LC-MS/MS as described

previously (Borchers et al., 1999). Briefly, the proteins were reduced, alkylated, and

digested with modified trypsin. We analyzed the peptide mixtures using a Waters Q-Tof

micro, hybrid quadropole orthogonal acceleration time-flight mass spectrometer (Waters,

Manchester, UK) with a Waters CapLC system configured with a PepMapTM _{C18 column.}

All data were acquired using Masslynx 4.0 software and then processed using the Proteinlynx

module. The processed data was searched against updated NCBInr and Sprot databases

using Mascot search engine. Mascot probability based Mowse individual ions scores > 46

were accepted as indicating identity or extensive homology (p<0.05). The MS/MS spectrum

scores between 20-45 were examined individually with the acceptance criteria beingthat the

parent and fragment ion masses were within the calibrated tolerance limits and that the

spectrum contained an extended series of consecutive y- or b- ions.

Western blot analysis

Small scale experiments analyzed by western blot were performed similarly to large

scale experiments with several exceptions. Lysates were generated from two 100mm dishes

of Calu-3 cells and CFTR affinity columns were generated with 2 nmoles of CFTR peptide

30

analyzed by western blot using anti-sera generated against NHERF-1 (Affinity bioreagents),

NHERF-3 (provided by David Silver, Columbia University), Ezrin (Santa Cruz), and Sorting

nexin 27 (provided by Wanjin Hong, Institute of Molecular and Cell Biology, Republic of

Korea).

Results

Our goal was to develop a biochemical approach to study protein interactions which

is MS compatible without requiring additional purification steps. To accomplish this, we

designed a protocol with special attention to lysate preparation, choice of affinity ligand, and

method of elution which allows us to maximally identify the proteins in our samples without

further purification. Our basic approach involves using small, highly purified peptides

instead of much larger fusion protein or antibodies to capture protein complexes. In addition,

we developed a novel elution protocol using formic acid which disassociates protein bound

to the beads, but does not add further contaminants to the sample. In this chapter we evaluate

our method by comparing our results with a sample prepared in parallel, but analyzed by a

more conventional approach utilizing SDS-PAGE separation and in-gel digestion (Figure

2.1A).

To test our method, we took advantage a well-studied protein interaction interface on

the C-terminus of CFTR. CFTR is an epithelial specific chloride channel mutated in cystic

fibrosis. The extreme C-terminus of CFTR contains a PDZ binding motif which terminates

in the sequence D-T-R-L (Wang et al., 1998). Multiple PDZ proteins interact with CFTR to

compartmentalize CFTR with regulatory factors and the cytoskeleton. Furthermore, PDZ

Figure 2.1. Strategy to purify PDZ proteins associated with CFTR

Work from our lab and others have described several PDZ proteins which associate with this

motif including the sodium hydrogen exchange regulatory factor-1 (NHERF-1 or EBP50)

(Short et al., 1998), NHERF-2 (E3KARP) (Sun et al., 2000b), NHERF-3 (PDZK1, CAP70,

NaPi CAP-1) (Wang et al., 2000), and the CFTR-associated ligand (CAL) (Cheng et al.,

2002). Studies of CFTR proteins lacking the PDZ binding motif have suggested that PDZ

proteins regulate multiple aspects of CFTR biology. However, it is not clear whether

additional, unidentified PDZ proteins may be involved in CFTR regulation. Thus, we sought

to identify the entire complement of PDZ proteins which associate with the C-terminus of

CFTR. To accomplish this, we probed mouse kidney cell lysates with a CFTR peptide

affinity column and analyzed the bound proteins by LC-MS/MS. The affinity column was

generated by coupling biotinylated, synthetic CFTR peptides, corresponding to the last 10

amino acids of CFTR (CFTR[1471-1480]), to streptavidin agarose beads. The peptides were

N-terminally biotinylated which contain a serine-glycine-serine-glycine (SGSG) linker

region followed by the CFTR sequence with a free C-terminal carboxylate group (Figure

2.1B). Control peptides were identical to CFTR[1471-1480] peptides, with the exception

that the D-T-R-L PDZ binding motif was mutated to glycines (CFTR[1471-1480/4G]).

To isolate protein associated with the CFTR C-terminus, detergent soluble mouse

kidney lysates were incubated with either CFTR[1471-1480] or CFTR[1471-1480/4G]

affinity matrices. We chose mouse kidney lysates for these studies as they are a rich source

of epithelia. The bound proteins were extensively washed, followed by a final wash in a low

salt buffer. For our in-solution approach, bound proteins were eluted using formic acid, were

subsequently diluted, lyophilized, and digested with trypsin in-solution (Figure 2.1A). We

34

disassociate proteins bound to the beads and it is volatile and removed from the sample

during lyophilization. Following digestion, the peptides were analyzed by LC-MS/MS using

a Waters Q-Tof micro. As a comparison, we prepared duplicate samples in which we used a

more conventional elution and analysis approach. We chose the following procedures as

they follow the standard protocol for MS analysis of proteins from acrylamide gels at the

UNC-Duke Michael Hooker proteomics center, which is similar to many other core

proteomics facilities. After incubating the CFTR affinity column with mouse kidney lysates,

the bound proteins were eluted with Laemmli sample buffer. The proteins were separated by

SDS-PAGE electrophoresis and stained with coomassie blue or silver stain (Invitrogen). All

visible bands were excised, digested in-gel with trypsin, and analyzed by MALDI-TOF/TOF

MS/MS as previously described (Parker et al., 2005).

By comparison, we identified more proteins with higher sequence coverages using

our in-solution method versus the conventional in-gel method. The silver stained, SDS

PAGE separated sample revealed multiple bands which specifically co-purify with the

CFTR[1471-1480] peptide, but not the CFTR[1471-1480/4G] mutant control (Figure 2.1C).

We successfully identified proteins from all visible bands as either NHERF-1 or NHERF-3,

both proteins known to associate directly with the CFTR COOH-terminus (Table 1). While

we did identify NHERFs-1 and -3, two known direct binding partners of CFTR, by this

conventional method, we did not identify NHERF-2, a previously identified binding partner.

Furthermore, we did not identify any indirect or novel binding partners. We also identified

1 and 3 by our in-solution method (Table 1). Moreover, we find

Table 1. A comparison of proteins identified by in-gel and in-solution approaches

Figure 2.2. The PDZ proteins identified by MS specifically associate with the CFTR C-terminus

NHERF-2. Our in-solution method also resulted in the identification of a novel

CFTR-associated protein, sorting nexin 27 (SNX27). Sorting nexins are proteins that regulate

intracellular trafficking events. SNX27 contains a PDZ binding motif, a lipid binding PX

domain, and a putative Ras-association/FERM domain. While the characterization of the

CFTR-SNX27 interaction is not described in this dissertation, our data suggest that SNX27

regulates the endocytic trafficking of CFTR. We confirmed our MS protein identifications

by western blot. Using antibodies specific to NHERF-1, NHERF-3, Ezrin, and SNX-27, we

find that each of these proteins specifically co-purifies with the CFTR[1471-1480] wild-type

peptide, but not the CFTR[1471-1480/4G] mutant (Figure 2.2).

Discussion

Here we describe a procedure which can be used to study protein-protein interactions

by MS without costly additional purification steps. Our method relies on using highly

purified peptides as affinity ligands, a low salt and detergent-free wash, and an acid elution.

Based on our MS data, our approach significantly improved our ability to identify proteins

associated with our affinity column.

The use of peptides in our approach is important for minimizing contaminants in the

samples. By comparison, traditional affinity ligands such as GST-fusion proteins or

antibodies are much larger and generate many more tryptic fragments in the sample. In

addition, the small size of the peptides tends to minimize non-specific protein binding.

Furthermore, peptides from commercial sources can be purchased at >95% purity, thus

offering a nearly homogenous affinity reagent. One limitation to using peptides is that they

40

domains. However, the peptide approach is well suited for studying protein-protein

interactions associated with the cytosolic tails of transmembrane proteins. This is

particularly useful in light of the observations that large transmembrane proteins are

notoriously difficult to study by MS as they are often endogenously expressed at low levels

and are only solubilized using high detergent concentrations which may disrupt protein

interactions. In addition, this approach could also be used to study other protein interactions

which involve linear sequences such as the SH3 or WW domains which interact with poly

proline motifs.

While we did not detect any background proteins using the in-gel procedure, we only

identified two direct binding proteins. In contrast, we identified more direct binding proteins,

indirectly associated proteins, and previously uncharacterized binding proteins using the

in-solution approach. We did detect a single background protein in the sample, propionyl

coenzyme A carboxylase, co-purified with both the wild-type CFTR[1471-1480] peptide and

the 4G mutant control. Propionyl coenzyme A carboxylase (PCC) is endogenously bound to

biotin and thus, likely associates with the streptavidin beads independently of the peptides

(Wood and Barden, 1977). It is also important to note that formic acid elution is stringent

enough to disassociate the biotinylated CFTR peptides from the streptavidin agarose. While

the peptide is separated from larger proteins by SDS-PAGE, the peptide was detected by MS

using the in-solution method. However, because of the small size of the peptide (yielding

only 2 tryptic fragments), the CFTR peptide is sufficiently separated from relevant peptides

in the sample by HPLC and did not interfere with the MS analysis.

Using our approach we identified more proteins and achieved higher sequence

identified a novel CFTR binding protein, SNX27. We predict that SNX27 interacts directly

with CFTR via its PDZ domain. However, it may also interact indirectly through

NHERFs-1, -2, or -3 via a PDZ or FERM domain interaction. Although, we do not describe the

characterization of the CFTR-SNX27 interaction in this dissertation, we find that SNX-27 is

highly concentrated in early endosomes where it likely regulates the endocytic sorting and

recycling of CFTR. In addition to improving our ability to correctly identify proteins using

MS data, the high sequence coverages improve the odds for the identification of

post-translational modifications, such as phosphorylation. Indeed, we identified one

phospho-peptide in our sample corresponding to a serine-phosphorylation at position 285 of NHERF-1

(data not shown). Previous studies have demonstrated that NHERF-1 is phosphorylated at

multiple sites (Fouassier et al., 2005; Reczek et al., 1997). However, serine 285 has not been

previously characterized.

In summary, we describe a novel approach for isolating and characterizing the

components of protein complexes. First, the use of peptides as affinity ligands results in

samples with low background due to proteins binding non-specific, especially compared to

larger affinity ligands. Secondly, a brief wash in a low salt buffer helps reduce residual salts

and detergents. Finally, the formic acid elution allows the purified proteins to be recovered

from the affinity matrix, without adding additional contaminants. Taken together, this

method provides a highly effective approach to the purification and preparation of samples

42

Chapter 3: CFTR is regulated by a direct interaction with the protein phosphatase

PP2A

Introduction

Cystic Fibrosis (CF) is an autosomal lethal disease characterized by abnormal ion

transport in epithelial tissues. CF is caused by mutations in the cystic fibrosis

transmembrane conductance regulator (CFTR), an apical membrane chloride channel. The

regulation of CFTR by the cAMP-dependent protein kinase (PKA) and other protein kinases

has been extensively documented. PKA can phosphorylate the CFTR regulatory domain (R

domain) on at least 11 serine residues (Chang et al., 1993; Cheng et al., 1991; Seibert et al.,

1995). In vivo, PKA phosphorylation increases CFTR open probability and the number of

channels in the plasma membrane (Ameen et al., 2003; Gadsby and Nairn, 1999). Work

from our lab and others have demonstrated that PKA and other regulatory proteins are

compartmentalized in close proximity to CFTR. The cellular machinery capable of

generating cAMP, including the adenosine receptor and membrane bound adenylate cyclase

are present with CFTR in apical membrane patches (Huang et al., 2001). A-kinase anchoring

proteins (AKAPs) target PKA to protein complexes with CFTR (Huang et al., 2000; Sun et

al., 2000a; Sun et al., 2000b) and the disruption of the PKA/AKAP interaction abolishes the

ability of PKA to activate CFTR in response to physiologic stimuli (Huang et al., 2000). In

addition, the phosphodiesterase PDE4D is also present with CFTR in patch preparations and

forms a cAMP diffusion barrier at the apical plasma membrane (Barnes et al., 2005). Other

kinase (AMPK) are found in protein complexes associated directly with CFTR (Hallows et

al., 2000; Liedtke et al., 2002).

Less is known about the ability of serine/threonine phosphatases to regulate CFTR

activity or how they are compartmentalized with CFTR. Work from many labs suggests that

multiple phosphatases including PP2A, PP2B, PP2C, and alkaline phosphatase are involved

in the deactivation of CFTR (Fischer et al., 1998; Hwang et al., 1993; Luo et al., 1998;

Reddy and Quinton, 1996; Travis et al., 1997). In vitro, PP2A and PP2C are most effective

in dephosphorylating purified CFTR and recombinant R domain. Furthermore, exogenous

PP2A and PP2C inactivate CFTR in excised membrane patches (Berger et al., 1993; Luo et

al., 1998). In human sweat ducts, cardiac myocytes, 3T3 fibroblasts, and Hi-5 insect cells,

inhibitors of PP2A increase CFTR channel activity (Berger et al., 1993; Hwang et al., 1993;

Reddy and Quinton, 1996; Yang et al., 1997). Likewise, PP2B inhibitors stimulate CFTR in

NIH 3T3 fibroblasts (Fischer et al., 1998). However, the contribution of PP2A and PP2B to

CFTR deactivation may vary in different cell types (Travis et al., 1997; Zhu et al., 1999). In

baby hamster kidney cells (BHK), PP2C co-immunoprecipitates with exogenously expressed

CFTR (Zhu et al., 1999). Additionally, PP2C overexpression in Fischer rat thyroid cells

(FRT) decreases CFTR chloride conductance (Travis et al., 1997). However, the ability of

endogenous PP2C to regulate CFTR in native epithelial tissues is unclear as no PP2C

inhibitors have been identified. To date, no single phosphatase has been demonstrated to be

both necessary and sufficient to completely down regulate CFTR channel activity, suggesting

that CFTR is dephosphorylated by multiple phosphatases.

Here, we present evidence for a direct interaction between CFTR and the